144 lines
3.6 KiB
JSON
144 lines
3.6 KiB
JSON
|
{
|
||
|
|
"best_global_step": 3170,
|
||
|
|
"best_metric": 76.40383058119879,
|
||
|
|
"best_model_checkpoint": "mbert-loraxs-qa-vanilla/run-2/checkpoint-3170",
|
||
|
|
"epoch": 6.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 3804,
|
||
|
|
"is_hyper_param_search": true,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"grad_norm": 7.677847385406494,
|
||
|
|
"learning_rate": 0.0002951562058189359,
|
||
|
|
"loss": 3.075869587317626,
|
||
|
|
"step": 634
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"eval_exact_match": 56.24,
|
||
|
|
"eval_f1": 70.50074993342362,
|
||
|
|
"eval_loss": 1.4242188930511475,
|
||
|
|
"eval_runtime": 7.8142,
|
||
|
|
"eval_samples_per_second": 161.884,
|
||
|
|
"eval_steps_per_second": 10.238,
|
||
|
|
"step": 634
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"grad_norm": 5.153384208679199,
|
||
|
|
"learning_rate": 0.00023614358062524137,
|
||
|
|
"loss": 1.4205032493037757,
|
||
|
|
"step": 1268
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"eval_exact_match": 59.52,
|
||
|
|
"eval_f1": 73.18016605585976,
|
||
|
|
"eval_loss": 1.3157330751419067,
|
||
|
|
"eval_runtime": 7.7666,
|
||
|
|
"eval_samples_per_second": 162.876,
|
||
|
|
"eval_steps_per_second": 10.3,
|
||
|
|
"step": 1268
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0,
|
||
|
|
"grad_norm": 2.3754422664642334,
|
||
|
|
"learning_rate": 0.00017713095543154682,
|
||
|
|
"loss": 1.1290527728829851,
|
||
|
|
"step": 1902
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0,
|
||
|
|
"eval_exact_match": 60.24,
|
||
|
|
"eval_f1": 74.7400274785974,
|
||
|
|
"eval_loss": 1.2541390657424927,
|
||
|
|
"eval_runtime": 7.8554,
|
||
|
|
"eval_samples_per_second": 161.037,
|
||
|
|
"eval_steps_per_second": 10.184,
|
||
|
|
"step": 1902
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0,
|
||
|
|
"grad_norm": 4.145329475402832,
|
||
|
|
"learning_rate": 0.0001181183302378523,
|
||
|
|
"loss": 0.9462081957314669,
|
||
|
|
"step": 2536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0,
|
||
|
|
"eval_exact_match": 62.32,
|
||
|
|
"eval_f1": 75.6887175752577,
|
||
|
|
"eval_loss": 1.28548264503479,
|
||
|
|
"eval_runtime": 7.7663,
|
||
|
|
"eval_samples_per_second": 162.882,
|
||
|
|
"eval_steps_per_second": 10.301,
|
||
|
|
"step": 2536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"grad_norm": 6.123648166656494,
|
||
|
|
"learning_rate": 5.9105705044157765e-05,
|
||
|
|
"loss": 0.8173495030929614,
|
||
|
|
"step": 3170
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"eval_exact_match": 62.48,
|
||
|
|
"eval_f1": 76.40383058119879,
|
||
|
|
"eval_loss": 1.3368667364120483,
|
||
|
|
"eval_runtime": 7.787,
|
||
|
|
"eval_samples_per_second": 162.449,
|
||
|
|
"eval_steps_per_second": 10.273,
|
||
|
|
"step": 3170
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.0,
|
||
|
|
"grad_norm": 3.7261476516723633,
|
||
|
|
"learning_rate": 9.307985046324058e-08,
|
||
|
|
"loss": 0.7344131710401469,
|
||
|
|
"step": 3804
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.0,
|
||
|
|
"eval_exact_match": 62.16,
|
||
|
|
"eval_f1": 76.11426067398186,
|
||
|
|
"eval_loss": 1.385920524597168,
|
||
|
|
"eval_runtime": 7.7794,
|
||
|
|
"eval_samples_per_second": 162.61,
|
||
|
|
"eval_steps_per_second": 10.284,
|
||
|
|
"step": 3804
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 500,
|
||
|
|
"max_steps": 3804,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 6,
|
||
|
|
"save_steps": 500,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": true
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 1.854297438697267e+16,
|
||
|
|
"train_batch_size": 16,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": {
|
||
|
|
"learning_rate": 0.0003186123281356725,
|
||
|
|
"lora_alpha": 64,
|
||
|
|
"lora_dropout": 0.0985087648472941,
|
||
|
|
"lora_r": 256,
|
||
|
|
"num_train_epochs": 6,
|
||
|
|
"weight_decay": 0.08602299412879012
|
||
|
|
}
|
||
|
|
}
|